A Folding-Docking-Affinity framework for protein-ligand binding affinity prediction.

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Accurate protein-ligand binding affinity prediction is vital for efficient drug discovery.
• Current docking-free methods often neglect explicit atom-level interactions when 3D binding structures are unavailable.
• Advancements in AI-based protein structure prediction offer new opportunities.

Purpose of the Study:

• To develop and evaluate a framework integrating protein folding, binding conformation determination, and affinity prediction.
• To assess the utility of utilizing predicted 3D protein-ligand binding structures for affinity prediction.
• To explore the potential of AI in enhancing binding affinity prediction accuracy.

Main Methods:

• The Folding-Docking-Affinity (FDA) framework was developed.
• FDA utilizes deep learning AI for protein folding and binding conformation prediction.
• Binding affinities are predicted directly from the generated 3D protein-ligand binding structures.

Main Results:

• The FDA framework demonstrates performance comparable to state-of-the-art docking-free methods.
• The study successfully integrates predicted protein structures into the binding affinity prediction pipeline.
• Experimental results validate the feasibility of the proposed approach.

Conclusions:

• The FDA framework provides a viable approach for binding affinity prediction using predicted structures.
• Integrating predicted binding conformations offers a promising avenue for improving accuracy.
• This work serves as a foundation for future structure-based affinity prediction methods.